Wireless structural control using stochastic bandwidth allocation and dynamic state estimation with measurement fusion
Wireless structural control using stochastic bandwidth allocation and dynamic state estimation with measurement fusion
复制标题
使用随机带宽分配和测量融合动态状态估计的无线结构控制
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
R. Andrew Swartz
中科院分区：
文献类型：
--
作者：
Benjamin D. Winter;R. Andrew Swartz
由于安装成本低，无线传感器网络在结构监测中越来越受欢迎；此外，将结构控制与无线数据采集相结合也可以产生优势。然而，这些系统的通信带宽有限，如果使用集中控制方法，随着控制网络中设备数量的增加，其有效性也会受到限制。在无线结构控制网络中配置数据的传统方法依赖于时间预算带宽或空间分散，其中网络被划分为更小的子网。这些方法很大程度上是静态的，并且通常不考虑任何数据质量测量来确定传输的优先级。本研究提出了一种无线结构控制网络中带宽分配的动态方法，该方法依赖于特定于应用的、自主的、控制器感知的、带有冲突检测协议的载波侦听多路访问。导出随机参数，以通过基于节点可观测性和输出估计误差的冲突检测算法策略性地改变载波侦听多路访问中的退避时间。受数据融合方法的启发，本文提出了两种不同的邻域状态估计方法，使用动态形式的仅测量融合。从竞争的无线介质接收到数据后，每个无线单元使用对应动态邻域的预先计算的静态卡尔曼增益矩阵来融合传入数据。每个无线单元都包含一个卡尔曼增益矩阵库，以容纳任何可能的通信数据集。给出了数值模拟和小规模实验室实验结果。
Wireless sensor networks are becoming more popular for structural monitoring because of their low installation costs; in addition, coupling structural control with wireless data acquisition can also yield advantages. However, these systems have limited communication bandwidth, limiting their effectiveness as the number of devices in control networks grows large if centralized control approaches are used. Traditional methods for collocating data in wireless structural control network rely on time‐budgeted bandwidth or spatial decentralization, where the network is divided into smaller subnetworks. These methods are largely static and typically do not take into account any measure of data quality to prioritize transmissions. This study presents a dynamic approach for bandwidth allocation in wireless structural control networks that relies on an application‐specific, autonomous, and controller‐aware, carrier sense multiple access with collision detection protocol. Stochastic parameters are derived to strategically alter back‐off times in the carrier sense multiple access with collision detection algorithm based on nodal observability and output estimation error. Inspired by data fusion approaches, this paper presents 2 different methods for neighborhood state estimation using a dynamic form of measurement‐only fusion. Upon receiving data from the contended wireless medium, each wireless unit fuses incoming data using a precalculated static Kalman gain matrix for the corresponding dynamic neighborhood. Onboard, each wireless unit contains a library of Kalman gain matrices, to accommodate any possible set of communicated data. Both numerical simulations and small‐scale laboratory experimental results are presented.